{
\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}
\begin{tabular}{l*{7}{c}}
\hline\hline
                &Business days         &Replied in time         &Complete answer         &Replied within 15 bd         &Replied within 25 bd         &Replied in 15 bd         &Replied in 25 bd         \\
\hline
\vspace{-6mm}   &                  &                  &                  &                  &                  &                  &                  \\
[1em]
1 day closer    &   -0.214         &    0.004         &   -0.001         &    0.024\sym{***}&    0.003         &    0.001         &   -0.007         \\
                &  (0.250)         &  (0.009)         &  (0.007)         &  (0.008)         &  (0.008)         &  (0.002)         &  (0.005)         \\
\hline
\addlinespace[3pt]Observations&      124         &      128         &      128         &      128         &      128         &      128         &      128         \\
\addlinespace[3pt]R-squared&    0.518         &    0.468         &    0.321         &    0.441         &    0.510         &    0.190         &    0.368         \\
\hline\hline
\multicolumn{8}{l}{\footnotesize  {\begin{minipage}{1.5\textwidth} \small   \vspace{0.2cm}  \textit{Notes:} * p $ < $ 0.10, ** p $ < $ 0.05, *** p $ < $ 0.01. We estimate the spillover effect of the redesigned notice on requests received with the old notices, which were the same as those the control group kept receiving. Robust standard errors are shown in parentheses. Only requests that were still pending when a redesigned notification arrived at the agency are included (Case II in Figure A3). Control variables include:  agency fixed effects, serial requesters, and number of requests received by the agency on the day request $ i $ arrived.     \end{minipage}  } }\\
\end{tabular}
}
